User Guide

This guide provides step by step directions for fitting semtree models to datasets with unmodeled covariates. The current version of semtree allows users to use either OpenMx or lavaan to define structures and fit models to datasets. Three subsections are detailed:

Using OpenMx with semtree

  • A guide to fitting a latent growth curve model to simulated five time point data with OpenMx, testing for homogenous subsets of cases based on unmodeled covariates, and plotting results of the fitted semtree.

Using lavaan with semtree

  • Another guide to fitting a latent growth curve model to simulated data with the lavaan R package, testing for homogenous subsets of cases based on unmodeled covariates, and plotting results of the fitted semtree.

Using OpenMx with semforest

  • This part of the guide continues with the latent growth curve model and simulated five time point data with OpenMx, and introduces a random forest algorithm to create semtrees with random subsets of data and unmodeled covariates, plotting variable importance measures, and plotting case proximity values.

Custom Stopping Rules

  • A guide to implementing a stopping rule not specifically provided in the control object of semtree. Provides an example of introducing a simple minimum ratio between the likelihood ratio between nested multi-group models and the degrees of freedom between nested multi-group models.